Development of an in-core fuel management tool for boiling water reactors

The in-core fuel management of a nuclear reactor is a challenging task due to the virtually infinite number of loading patterns one could theoretically adopt. The ROSA (Reloading Optimization by Simulated Annealing) code is an optimization tool that has been successfully used in the last two decades to facilitate the core design of several Pressurized Water Reactors (PWRs). It is designed to perform a stochastic search for an optimal Loading Pattern (LP) using a simulated annealing algorithm. This corresponds to performing a depletion calculation for each one of the hundreds of thousands of unique LPs generated during the stochastic search. Therefore, speed is one of the most important requirements that the solvers used by the depletion tool must fulfill. ROSA's depletion analysis tool makes use of a particularly fast nodal method (known as the kernel method) for the evaluation of the power distribution associated with a particular LP. One of the strongest assumptions behind the kernel method is that the neutron migration length does not change considerably between the point where a neutron is generated and the point where the same neutron is absorbed. Although strong, this assumption is quite compatible with the neutronic characteristics of PWRs cores. In this paper we give an overview of the work done in order to develop a version of ROSA capable of performing the core design of Boiling Water Reactors (BWRs). We focus the discussion on the development of the depletion analysis tool by outlining the modifications of the kernel methods implemented in order to make the solver accurate for BWR cores. An improvement of the definition of the transport kernel is necessary to take the strong anisotropies characterizing the neutronic problem into account. These anisotropies arise due to the presence of strong changes in the moderator density and due to the presence of control blades. Furthermore, we are going to discuss how the boundary conditions are adopted by the neutronic solver in the form of albedos, which are used as a calibration parameter to tune the solution using the information obtained from an external license code. We then analyze the accuracy of ROSA-BWR when predicting power distributions and safety parameters by benchmarking it against the core follow calculations obtained by using the license code MICROBURN-B2 for three cycles of the Browns Ferry 3 nuclear power plant. This benchmark is also discussed in terms of computational performances and potential gains achievable through parallelization. The final section of the paper contains a description of the optimization capabilities that are currently implemented in ROSA-BWR and an overview of the preliminary applications of the code to real life core design studies. (author)